Homeostatic regulation of ribosomal proteins by ubiquitin-independent cotranslational degradation

Significance

Ribosomes are macromolecular machines that perform protein biosynthesis. The concentration of ribosomes is critical for proper protein production to support normal cell function. Therefore, it is imperative to understand the mechanism that controls the levels of ribosomal proteins. Here, we demonstrate that while mature ribosomal proteins are stable, nascent ribosomal proteins are subject to high-level cotranslational protein degradation (CTPD). The rate of CTPD is determined by cotranslational folding potential and expression levels of ribosomal proteins. Unlike conventional posttranslational degradation, CTPD of all ribosomal proteins tested is ubiquitin- independent. This study sheds light on the mechanism regulating homeostasis of ribosomal proteins.

Abstract

Ribosomes are the workplace for protein biosynthesis. Protein production required for normal cell function is tightly linked to ribosome abundance. It is well known that ribosomal genes are actively transcribed and ribosomal messenger RNAs (mRNAs) are rapidly translated, and yet ribosomal proteins have relatively long half-lives. These observations raise questions as to how homeostasis of ribosomal proteins is controlled. Here, we show that ribosomal proteins, while posttranslationally stable, are subject to high-level cotranslational protein degradation (CTPD) except for those synthesized as ubiquitin (Ub) fusion precursors. The N-terminal Ub moiety protects fused ribosomal proteins from CTPD. We further demonstrate that cotranslational folding efficiency and expression level are two critical factors determining CTPD of ribosomal proteins. Different from canonical posttranslational degradation, we found that CTPD of all the ribosomal proteins tested in this study does not require prior ubiquitylation. This work provides insights into the regulation of ribosomal protein homeostasis and furthers our understanding of the mechanism and biological significance of CTPD.

Ribosomes are the cellular organelles responsible for protein biosynthesis. The abundance of ribosomes is tightly linked to protein production required for normal cell function, and therefore, ribosome biogenesis is highly regulated (1–3). Ribosomes are composed of two subunits, the small and the large subunit. The small subunit decodes messenger RNA (mRNA), whereas the large subunit catalyzes peptide bond formation. In eukaryotic cells, the small subunit (40S) contains an 18S ribosome RNA (rRNA) and 33 proteins, and the large subunit (60S) consists of three rRNAs (5S, 5.8S, and 28S) and 46 proteins. Coordinated expression of ribosomal proteins, which have a relatively fixed stoichiometry, is conceivably important for ribosome biogenesis. Notably, ribosomal genes are actively transcribed, and ribosomal mRNAs are rapidly translated (4–6). As such, ribosomal proteins are often expressed at high levels beyond that required for the typical rate of ribosome assembly (7–9). While overproduction may ensure that ribosomal protein levels are never rate limiting for ribosome biogenesis, it could have severe impacts. For example, like other RNA-binding proteins, excess ribosomal proteins that are not assembled into ribosomes tend to form toxic aggregates in the cell. Moreover, because of the trade-off between translation speed and accuracy (10, 11), accelerating translation of ribosomal mRNAs increases the chance of incorporating wrong amino acids into ribosomal proteins. To restrict accumulation of excess and/or erroneous ribosomal proteins is critical for cellular homeostasis. Sung et al. showed that overexpressed ribosomal protein Rpl26, which fails to assemble into ribosomes, is degraded by the Ub-proteasome system in yeast and human cells (12, 13). However, systematic analysis of protein stability in yeast cells revealed that endogenously expressed ribosomal proteins including Rpl26 have relatively long half-lives (14). This report indicates that mature ribosomal proteins are rather stable under normal conditions.

It is well documented that a substantial fraction of nascent proteins is degraded by the proteasome during translation, a process known as cotranslational protein degradation (CTPD) (15–17). The term CTPD is also used to describe degradation of nascent proteins that have just been released from ribosomes but not yet folded. Interestingly, nascent polypeptides bearing mutations and unable to achieve a folded conformation are prone to CTPD (16). This observation suggests that CTPD is an important mechanism preventing accumulation of misfolded, defective, toxic proteins in the cell. Could CTPD play a role in regulating ribosomal protein homeostasis? Unfortunately, this question has never been raised before. Part of the reason, at least, is due to the high abundance of ribosomal proteins. In addition, mature ribosomal proteins have long half-lives (14). It is often assumed that CTPD is directly related to the stability of mature proteins. Our recent study disproved this assumption. In the course of identifying proteins subject to CTPD using a quantitative proteomic approach, we found that there is no correlation between CTPD and the half-lives of mature proteins (18). This finding promoted us to assess CTPD of ribosomal proteins.

Here, we compiled the CTPD data of 62 ribosomal proteins acquired from the proteomic analysis. We found that all ribosomal proteins identified are subject to substantial CTPD except for L40 (also named Rpl40). Interestingly, L40 is synthesized as a Ub-fusion precursor (19). We demonstrated that the N-terminal Ub protects L40 from CTPD. We further showed that cotranslational folding efficiency and expression level are two determining factors for CTPD of ribosomal proteins. Unlike conventional posttranslational degradation, CTPD of ribosomal proteins tested does not require prior iquitylation. These data provide insights into homeostatic regulation of ribosomal proteins and add to our understanding of the biological significance and mechanism of CTPD.

Results

Ribosomal Proteins Are Subject to CTPD.

We recently identified that CTPD is impaired in the yeast mutant srp1-49 expressing a defective nuclear import factor Srp1 (also known as importin α, karyopherin α, or Kap60) (17). Taking advantage of this yeast mutant and the stable isotope labeling by amino acids in cell culture (SILAC) and puromycin-associated nascent chain proteomics (PUNCH-P) biochemical approaches (20, 21), we launched a quantitative proteomic analysis to identify nascent proteins subject to CTPD (18). Wild-type (WT) and srp1-49 cells were cultured in light (L-lysine) and heavy (L-lysine-[¹³C₆ ¹⁵N₂]) media, respectively. In label swap experiments, WT cells were grown in heavy medium, whereas srp1-49 in light medium. The SILAC ratios of WT versus srp1-49, reflecting the difference in abundance of individual ribosome-bound nascent chains between WT and srp1-49 cells, were used to calculate the rate of CTPD, i.e., the percentage of a nascent polypeptide degraded cotranslationally in WT cells. Our proteomic analysis acquired 1,422 nascent proteins, covering a significant portion of the yeast proteome. Approximately, 21% of the proteins undergo high-level CTPD, with >30% nascent chains degraded. Another 40% of the proteins are degraded at a medium level, with 10 to 30% nascent chains degraded. These data underscore the importance of CTPD in regulating protein homeostasis. Interestingly, bioinformatic analysis revealed that there is no correlation between CTPD and the half-lives of mature proteins. In the current study, we compiled the CTPD data of ribosomal proteins acquired from the proteomic analysis. Sixty-two out of 79 ribosomal proteins were identified, including 26 small subunit and 36 large subunit ribosomal proteins (SI Appendix, Table S1). Most ribosomal proteins are encoded by two genes producing identical or nearly identical proteins known as paralogs. Including the paralogous proteins, we actually obtained CTPD data for more than 62 ribosomal proteins. We found that all ribosomal proteins identified are subject to CTPD. Most of them have a CTPD rate over 30 to 50% (SI Appendix, Table S1). The ribosomal protein L40 has the lowest CTPD rate (13.6%). Interestingly, L40 is synthesized as a precursor with a Ub moiety at the N terminus (19). The N-terminal Ub is cleaved by Ub-specific proteases (UBPs), releasing L40 from the fusion protein. The relatively lower level CTPD of L40, as compared to other ribosomal proteins, suggests that the N-terminal Ub may protect L40 from CTPD. This was confirmed as described below. In summary, all ribosomal proteins identified by our proteomic analysis are subject to high-level CTPD except for L40.

The N-terminal Ub Protects L40 and Rps31 from CTPD.

L40 consisting of 52 amino acids is encoded by two paralogous genes UBI1 and UBI2. These two paralogous genes share 93% identity of DNA sequence, and their protein products Ub-L40A and Ub-L40B are identical. Here, we examined whether the N-terminal Ub protects L40A from CTPD by utilizing a pair of yeast strains, UBI1-HA and ubi1Δub-HA, generated by Martín-Villanueva et al. (SI Appendix, Fig. S1A, also see ref. 22). UBI1-HA is the parental strain with a hemagglutinin (ha) epitope tag attached to the C terminus of L40A, producing an ha-tagged precursor protein (Ub-L40Aha). In the ubi1Δub-HA mutant, the UBI1 locus was replaced by a ubi1Δub-HA allele, which is deleted of the UB sequence and therefore solely expresses ha-tagged L40A (L40Aha). Both UBI1-HA and ubi1Δub-HA alleles are driven by the UBI1 promoter in the native locus. Early study and our own observation showed that the growth rates of UBI1-HA and ubi1Δub-HA are similar. qRT-PCR analysis showed that the UBI1-HA transcript abundance was only ~50% of ubi1Δub-HA (SI Appendix, Fig. S1B). This is likely because the UBI1-HA allele is much larger and needs more time to transcribe than the ubi1Δub-HA counterpart. (The UB sequence includes an intron in addition to the Ub coding sequence). Despite the lower transcript level, the L40Aha protein level was higher in UBI1-HA than ubi1Δub-HA (SI Appendix, Fig. S1C). Metabolic labeling with ³⁵S-methionine/cysteine (Met/Cys) showed that protein synthesis rates were similar between these two strains (SI Appendix, Fig. S1D). Thus, the higher steady-state level of L40Aha in UBI1-HA was not caused by elevated transcription or translation, but instead due to reduction of L40Aha degradation. Cycloheximide (CHX)-chase analysis showed that L40Aha is posttranslationally stable (SI Appendix, Fig. S1E). Together, these results indicate that the higher steady-state level of L40Aha in UBI1-HA results from less CTPD and that the N-terminal Ub protects L40 from CTPD.

Rps31 is the other ribosomal protein synthesized as a Ub fusion precursor in eukaryotic cells (19). Although our proteomic analysis did not obtain Rps31, we wanted to examine whether the N-terminal Ub also protects Rps31 from CTPD. Rsp31 is a quasi-essential ribosomal protein of 76 amino acids. Different from L40, Rps31 is encoded by only one gene (UBI3) that contains no introns. Previous studies showed that deletion of the Ub coding sequence from UBI3 caused slow cell growth (19, 23, 24). It would be complicated to compare CTPD of Rps31 between UBI3 and ubi3Δub strains because the slow growth phenotype of ubi3Δub may be associated with compromised protein synthesis. We constructed two plasmids that express C-terminally ha-tagged Ub-Rps31 (UBI3-HA) or Rps31 (ubi3Δub-HA) from the copper-inducible CUP1 promoter in a low-copy-number vector (Fig. 1A). These two plasmids (p314CUP1UBI3-HA and p314CUP1ubi3Δub-HA) and a void control vector were transformed into a WT yeast strain. We found that the steady-state level of Rps31ha was much higher in the UBI3-HA transformant than the ubi3Δub-HA counterpart (Fig. 1B). Consistent with early reports (19, 23), no Ub-Rps31ha fusion protein was detected by western blotting due to rapid removal of the N-terminal Ub. The transcript levels of UBI3-HA and ubi3Δub-HA were comparable (Fig. 1C), indicating that the higher steady level of Rps31ha in the UBI3-HA transformant was not caused by higher transcription activity. CHX-chase analysis showed that mature Rps31ha was stable regardless of the presence or absence of N-terminal Ub (Fig. 1D). To further demonstrate the effect of the N-terminal Ub on CTPD of Rps31ha, we performed a short-pulse labeling assay. If Rps31ha is subject to higher-level CTPD than Ub-Rps31ha, the yield of nascent Rps31ha protein should be lower in the ubi3Δub-HA transformant than the UBI3-HA counterpart. We labeled the UBI3-HA and ubi3Δub-HA transformants with ³⁵S-Met/Cys for only 1 or 2.5 min. ³⁵S-labeled (nascent) Rps31ha was immunoprecipitated by an anti-ha antibody, followed by SDS-PAGE and autoradiography (Fig. 1E). The amount of ³⁵S-labeled Rps31ha in the ubi3Δub-HA transformant was only 25 to 35% of that in the UBI3-HA counterpart (Fig. 1F). These two transformants had similar protein synthesis rates (SI Appendix, Fig. S2). Together, these observations indicate that the N-terminal Ub protects Rps31 from CTPD.

Fig. 1.

N-terminal Ub protects Rps31 from CTPD. (A) Schematic of UBI3-HA and ubi3Δub-HA alleles encoding C-terminally ha tagged Ubi3 (Ub-Rps31ha) and Ub-less Ubi3 (Rps31ha). (B) Western blotting analysis of Rps31ha in yeast cells transformed with p314CUP1UBI3-HA (lane 2), p314CUP1ubi3Δub-HA (lane 3), or a void control vector (lane 1). Rps31ha and internal control phosphoglycerate kinase (Pgk1) are indicated with arrows. (C) Comparison of the abundances of UBI3-HA and ubi3Δub-HA transcripts. Cells transformed with p314CUP1UBI3-HA, p314CUP1ubi3Δub-HA, or a control vector were analyzed by qRT-PCR with oligo primers corresponding to the coding sequences of Rps31 and the ha tag. Data are mean ± SD of three independent experiments. (D) Rps31 is posttranslationally stable. CHX chase was performed to measure the stability of mature Rps31ha protein in cells expressing UBI3-HA or ubi3Δub-HA. (E) Short-pulse labeling of cells expressing UBI3-HA and ubi3Δub-HA. Cells were labeled with ³⁵S-Met/Cys for 1 or 2.5 min. Cell extracts were subjected to IP with an anti-ha antibody, followed by SDS-PAGE and autoradiography. Rps31ha is marked by an arrow. (F) ³⁵S-labeled Rps31ha from (E) was quantified by a PhosphorImager. The amount of Rps31ha in cells expressing ubi3Δub-HA was normalized against that in cells expressing UBI3-HA, which was set at 1.0.

CTPD of Rpl8A Is Impaired by Attachment of an N-terminal Ub.

We next tested whether adding a Ub moiety to the N termini of other ribosomal proteins could reduce CTPD. To this end, we constructed a UB vector that can be used to produce Ub-fusion proteins without altering the N-terminal sequences of the fusion partners after removal of the Ub moiety. This is important because a subtle change of the N-terminal amino acid sequence may affect CTPD. Briefly, we took advantage of the ligation of DNA fragments generated by the SfoI and SmaI restriction enzymes, which results in a GGCGGG linker between the Ub coding sequence (codons 1 to 74) and the start codon of your favorite gene (YFG) (Fig. 2A). The GGCGGG linker encodes the last two residues (Gly-Gly) of Ub. UBPs precisely cleave the N-terminal Ub, leaving your favorite protein (YFP) intact. The backbone of the UB vector is a CEN-based low-copy-number plasmid such that the variation of copy numbers of YFG among transformants is minimal. As a proof of concept, we examined whether adding an N-terminal Ub to Rpl8A could reduce its CTPD. Rpl8A is one of two paralogous Rpl8 ribosomal proteins. Both paralogs (Rpl8A and Rpl8B) are subject to high-level CTPD (SI Appendix, Table S1). Another reason to choose Rpl8A is because the RPL8A gene does not contain introns and therefore is relatively straightforward for molecular cloning and truncation-functional analysis. The CUP1 promoter was used to express C-terminally ha-tagged Rpl8A and Ub-Rpl8A. We transformed yeast cells with plasmids p314CUP1RPL8Aha or p314CUP1UB-RPL8Aha, controlled by a void vector. Western blotting showed that the Rpl8Aha protein level was much higher in the p314CUP1UB-RPL8Aha transformant than the p314CUP1RPL8Aha counterpart (Fig. 2B). No Ub-Rpl8Aha precursor protein was detected, indicative of rapid cleavage of the N-terminal Ub. qRT-PCR analysis revealed that the UB-RPL8Aha transcript level was slightly lower than RPL8Aha (Fig. 2C). Apparently, placing the Ub coding sequence to the 5′ end did not increase the transcription level of RPL8A. CHX chase assay showed that Rpl8Aha was posttranslationally stable in both p314CUP1RPL8Aha and p314CUP1UB-RPL8Aha transformants, whereas the Rpl8Aha protein level was much higher in the latter (Fig. 2D). These results demonstrate that placing a Ub moiety to the N terminus inhibits CTPD of Rpl8A.

Fig. 2.

Addition of Ub to the N terminus protects Rpl8A from CTPD. (A) Diagram of the UB vector used for production of Ub-fusion proteins. The backbone of the UB vector is a CEN-based low-copy-number plasmid. (B) Western blotting analysis of Rpl8Aha in yeast cells transformed with p314CUP1RPL8Aha (lane 2), p314CUP1UB-RPL8Aha (lane 3), or a void control vector (lane 1). Rpl8Aha and Pgk1 are marked with arrows. (C) Comparison of the levels of RPL8Aha and UB-RPL8Aha transcripts. qRT-PCR was performed to measure the abundances of RPL8Aha and UB-RPL8Aha transcripts in cells transformed with p314CUP1RPL8A and p314CUP1UB-RPL8A. Oligo primers specific to RPL8A and the C-terminal ha tag coding sequences were used. Shown are the results of three independent experiments. (D) CHX chase was performed to measure the posttranslational stability of Rpl8Aha in cells transformed with p314CUP1RPL8Aha and p314CUP1UB-RPL8Aha.

Inefficient Cotranslational Folding Leads to CTPD of Rpl8A.

To investigate the mechanism underlying CTPD of Rpl8A, we decided to delineate the sequence or degron that directs CTPD of Rpl8A. Of note, while nascent Rpl8A is subject to CTPD, the mature protein is stable. These observations suggest that the CTPD degron is exposed on nascent chains but shielded in the mature protein. To evaluate this model, we assessed the stability of several fragments of Rpl8A including amino acids 1 to 100, 1 to 158, and 1 to 200 (Fig. 3A), which resemble nascent chains of different lengths. An ha tag was added C-terminally to these fragments expressed from the CUP1 promoter in a low-copy-number vector. We compared the steady-state levels of these fragments with full-length Rpl8A. Unlike Rpl8Aha, the fragments Rpl8A_1-100ha, Rpl8A_1-158ha, and Rpl8A_1-200ha were virtually undetected by western blotting (Fig. 3B). qRT-PCR analysis revealed no significant difference in the abundances of transcripts encoding Rpl8Aha and the truncated fragments (Fig. 3C). These results suggest that Rpl8A_1-100ha, Rpl8A_1-158ha, and Rpl8A_1-200ha proteins are extremely unstable. We suspected that the CTPD degron may be continuously exposed on the Rpl8A fragments due to loss of interaction with other domains in the C-terminal region, thereby signaling not only CTPD but also posttranslational degradation. To test this possibility, we fused Rpl8A_1-100, Rpl8A_1-158, Rpl8A_1-200, and Rpl8A to the N terminus of dihydrofolate reductase (DHFR) (Fig. 3A). DHFR is a stable protein widely used as a reporter to assess degron activity. An ha tag was added to the C terminus of DHFR for detection. We observed that the steady-state levels of Rpl8A_1-100-DHFRha, Rpl8A_1-158-DHFRha, and Rpl8A_1-200-DHFRha were much lower than that of Rpl8A-DHFRha (SI Appendix, Fig. S3A). CHX-chase analysis showed that Rpl8A_1-100-DHFRha, Rpl8A_1-158-DHFRha, and Rpl8A_1-200-DHFRha were rapidly degraded (Fig. 3D and SI Appendix, Fig. S3B). In contrast, Rpl8A-DHFRha remained stable during CHX chase. Two conclusions are drawn from these observations. First, the N-terminal region of Rpl8A contains a CTPD degron, which lies within the first 100 amino acids. Second, the C-terminal region including amino acids 201 to 256 is required to conceal the CTPD degron in the mature protein.

Fig. 3.

The N-terminal region of Rpl8A harbors a CTPD degron. (A) Schematic of truncated Rpl8A and DHFR fusion proteins. All proteins were expressed from the CUP1 promoter in a CEN-based low-copy-number plasmid. (B) Western blotting to compare the steady-state levels of full-length and N-terminal fragments of Rpl8A. (C) qRT-PCR analysis to compare the levels of transcripts encoding full-length and truncated Rpl8A proteins. (D) CHX chase analysis to compare the posttranslational stability of DHFR fusion proteins with Rpl8A or N-terminal fragments. (E) Western blotting to compare the steady-state levels of full-length and C-terminally truncated Rpl8A proteins. The relative ratios of full-length and truncated Rpl8A proteins versus Pgk1 were shown. The protein bands were quantified by the software ImageJ. (F) CHX chase analysis to measure the posttranslational stability of full-length and C-terminally truncated Rpl8A proteins.

We further dissected the C-terminal region required to mask the CTPD degron. We generated several mutants with smaller C-terminal deletions, including Rpl8A_1-245, Rpl8A_1-230, and Rpl8A_1-215, which were deleted of 11, 26, and 41 amino acids from the C terminus (Fig. 3A). These mutants were expressed from the CUP1 promoter in a low-copy-number vector, and their steady-state levels were measured by western blotting, controlled by Rpl8Aha and Rpl8A_1-200ha. Unlike Rpl8A_1-200ha, which was virtually absent, Rpl8A_1-215ha, Rpl8A_1-230ha, and Rpl8A_1-245ha were readily detected (Fig. 3E). However, the steady-state levels of Rpl8A_1-215ha, Rpl8A_1-230ha, and Rpl8A_1-245ha were only 23%, 55%, and 49% of that of Rpl8A-ha. The difference in protein expression levels did not result from transcription because the abundances of transcripts encoding Rpl8Aha, Rpl8A_1-245ha, Rpl8A_1-230-ha, Rpl8A_1-215-ha, and Rpl8A_1-200-ha were comparable (Fig. 3C). CHX chase analysis showed that Rpl8A_1-245-ha, Rpl8A_1-230-ha, and Rpl8A_1-215-ha were not degraded during the chase period (Fig. 3F). Therefore, mature Rpl8A_1-215ha, Rpl8A_1-230ha, and Rpl8A_1-245ha proteins are stable, and their lower steady-state levels, as compared to Rpl8Aha, result from higher-level CTPD. This truncation-functional analysis demonstrates the requirement of the very C-terminal domain of Rpl8A for nascent polypeptide folding. A small deletion from the C terminus (e.g., 11 amino acids) readily compromises the folding efficiency, leading to more nascent polypeptide degradation. The truncated proteins become stable once folding is completed. These observations reiterate that the N-terminal region of Rpl8A cannot fold properly on its own, leaving the nascent chains vulnerable to CTPD until completion of protein synthesis. These data indicate that inefficient cotranslational folding contributes to high-level CTPD of Rpl8A.

Protein Expression Level Is a Crucial Determinant for CTPD of Rps12.

We next examined the effect of adding an N-terminal Ub on CTPD of Rps12, a small subunit ribosomal protein with a high CTPD rate (SI Appendix, Table S1). Using the UB vector, we constructed plasmid p314CUP1UB-RPS12Aha expressing C-terminally ha-tagged Ub-Rps12 from the CUP1 promoter (Fig. 4A). The plasmid p314CUP1RPS12ha expressing C-terminally ha-tagged Rps12 was derived from the same vector backbone without the N-terminal Ub coding sequence. These two plasmids, controlled by a void vector, were transformed into a WT yeast strain. The steady-state level of Rps12ha was measured by western blotting with an anti-ha antibody (Fig. 4B). Surprisingly, the Rps12ha protein level was relatively lower in the pCUP1UB-RPS12ha transformant than the pCUP1RPS12ha counterpart. qRT-PCR analysis showed that the abundances of UB-RPS12ha and RPS12ha transcripts were comparable (Fig. 4C). Why adding an N-terminal Ub did not increase the steady-state level of Rps12, unlike other ribosomal proteins tested? We suspected that Rps12 might not be subject to CTPD under the experimental conditions. Note that ribosomal genes are highly expressed, whereas the activity of CUP1 promoter is relatively moderate. It is possible that nascent Rps12 proteins escape from CTPD when expressed at a moderate level but are prone to CTPD when expressed at a high level. To test this possibility, we replaced the CUP1 promoter with the stronger galactose-inducible GAL1 promoter to drive the expression of Rps12ha and Ub-Rps12ha. Plasmids p414GAL1RPS12ha and p414GAL1UB-RPS12ha, controlled by the empty vector p414GAL1 (25), were transformed into yeast cells. Remarkably, the Rps12ha steady-state level was much higher in the p414GAL1UB-RPS12ha transformant than the p414GAL1RPS12ha counterpart (Fig. 4D). On the other hand, the UB-RPS12ha transcript level was slightly lower than RPS12ha (Fig. 4E). Thus, the increase in Rps12ha steady-state level was not caused by transcription alteration. We then examined the posttranslational stability of Rps12ha expressed from the GAL1 promoter using CHX-chase analysis. As shown in Fig. 4F, Rps12ha was stable during CHX chase regardless of the N-terminal Ub. This observation excludes the possibility that the N-terminal Ub increases Rps12ha steady-state level through inhibiting posttranslational degradation, which might result from GAL1 promoter-induced overexpression. Together, these results demonstrate that CTPD of Rps12 is determined by its expression level. Rps12 is subject to CTPD when expressed at a high level but resistant to CTPD when its expression level is moderate. Adding an N-terminal Ub minimizes CTPD of overexpressed Rps12.

Fig. 4.

CTPD of Rps12 is determined by protein expression level. (A) Schematic of Rps12ha and Ub-Rps12ha proteins. (B) Western blotting of Rps12ha in cells transformed with p314CUP1RPS12ha (lane 2), p314CUP1UB-RPS12ha (lane 3), or a control vector (lane 1). (C) qRT-PCR analysis to compare the levels of transcripts encoding Rps12ha and Ub-Rps12ha from cells as in (B). Oligo primers specific to the coding sequences of Rps12 and the ha tag were used. (D) Western blotting of Rps12ha expressed from the GAL1 promoter in cells transformed with p414GAL1RPS12ha or p414GAL1UB-RPS12ha. The transformant with a void vector served as a control. (E) qRT-PCR to compare the levels of RPS12ha and UB-RPS12ha transcripts in cells as in (D). (F) CHX chase assay was performed to measure the posttranslational stability of Rps12ha in cells overexpressing Rps12ha or Ub-Rps12ha from the GAL1 promoter.

CTPD of Ribosomal Proteins Does Not Require Prior Ubiquitylation.

To further understand the underlying mechanism, we went on to examine whether CTPD of ribosomal proteins needs prior ubiquitylation. To this end, we took advantage of the temperature-sensitive E1 Ub-activating enzyme mutant uba1-204 that exhibits rapid and drastic inhibition of Ub-dependent protein degradation at nonpermissive temperature (26). We first validated the effect of shifting uba1-204 to nonpermissive temperature on Ub-dependent protein degradation using Rpn4_172-229-DHFRha as the substrate. Our early study demonstrated that proteasomal degradation of Rpn4_172-229-DHFRha is Ub dependent (27). We transformed WT and uba1-204 cells with the plasmid p314CUP1RPN4(172-229)-DHFRha expressing Rpn4_172-229-DHFRha from the CUP1 promoter. The transformants were initially induced with CuSO₄ at permissive temperature (28 °C) for 1 h. Half of each culture was then shifted to nonpermissive temperature (37 °C) for additional 2 h induction, whereas the other half was continuously induced at the permissive temperature. Cell extracts were analyzed by western blotting with an anti-ha antibody. Degradation of Rpn4_172-229-DHFRha was inhibited in uba1-204 at 37 °C, but as efficient as in WT cells at 28 °C (Fig. 5A). This experiment confirmed the defect of Ub-dependent protein degradation in uba1-204 at nonpermissive temperature.

Fig. 5.

Ub-independent CTPD of ribosomal proteins. (A) Western blotting to compare the steady-state levels of Rpn4_172-229-DHFRha in WT and uba1-204 cells at permissive and nonpermissive temperatures. The transformants with plasmid p314CUP1RPN4(172-229)-DHFRha were first induced with CuSO₄ (0.1 mM) at 28 °C for 1 h. Half of each culture was then shifted to 37 °C for additional 2 h induction, whereas the other half was continuously induced at 28 °C. (B) Comparison of the steady-state levels of Rpl8Aha in WT and uba1-204 cells transformed with plasmids p314CUP1RPL8Aha (lanes 1, 3, 5, and 7) or p314CUP1UB-RPL8Aha (lanes 2, 4, 6, and 8). CuSO₄ induction was performed as in (A). (C) Comparison of the steady-state levels of Rps31ha in WT and uba1-204 cells transformed with plasmids p314CUP1UBI3-HA (lanes 1, 3, 5, and 7) or p314CUP1ubi3Δub-HA (lanes 2, 4, 6, and 8). CuSO₄ induction was conducted as in A. WT cells carrying a void vector served as a control (lane 0). (D) Comparison of the steady-state levels of Rps12ha in WT and uba1-204 cells transformed with plasmids p414GAL1RPS12ha (lanes 1, 3, 5, and 7) or p414GAL1UB-RPS12ha (lanes 2, 4, 6, and 8). The transformants were grown to log phase in a medium with raffinose as the carbon source and transferred to a medium containing galactose. Initial galactose induction took place at 28 °C for 1 h. The cultures were then split into two parts: one shifted to 37 °C, whereas the other kept at 28 °C for additional 2 h induction. WT cells containing a void vector served as a control (lane 0).

We transformed plasmids p314CUP1RPL8Aha and p314CUP1UB-RPL8Aha into WT and uba1-204 cells. The transformants were first induced with CuSO₄ at 28 °C for 1 h. Half of each culture was then shifted to 37 °C, whereas the other half was maintained at 28 °C, for additional 2 h induction. Western blotting was performed to measure the steady-state levels of Rpl8Aha in these transformants. The difference in Rpl8Aha steady-state levels between cells expressing Rpl8Aha and Ub-Rpl8Aha reflects the level of CTPD. We found that CTPD of Rpl8Aha in uba1-204 was as efficient as in WT cells at both 28 °C and 37 °C (Fig. 5B). Apparently, inactivation of the E1 enzyme exhibits no effect on CTPD of Rpl8Aha. We also transformed plasmids p314CUP1UBI3-HA and p314CUP1ubi3Δub-HA into WT and uba1-204 cells to compare CTPD of Rps31ha at 28 °C and 37 °C. Like Rpl8A, CTPD of Rps31ha in uba1-204 was not affected by inactivation of the E1 enzyme at 37 °C (Fig. 5C). We next examined whether loss of E1 activity would inhibit CTPD of Rps12ha overexpressed from the GAL1 promoter. Plasmids p414GAL1RPS12ha and p414GAL1UB-RPS12ha were transformed into WT and uba1-204 cells. The transformants grown to the log phase in a medium with raffinose as the carbon source were transferred to a medium containing galactose. The initial galactose induction took place at 28 °C for 1 h. The cultures were then split into two parts: one shifted to 37 °C, whereas the other kept at 28 °C, for additional 2 h induction. Western blotting analysis showed that shifting uba1-204 cells to 37 °C did not impair CTPD of Rps12ha (Fig. 5D). These results conclude that CTPD of Rpl8A, Rps31, and Rps12 is Ub-independent.

CTPD of Rpl8A and Rps31 Is Not Impaired in 19S Proteasome Mutants.

The demonstration of Ub-independent CTPD of ribosomal proteins prompted us to assess the requirement for proteasomes. The 26S proteasome is composed of a 20S proteasome (core) and two 19S proteasomes (regulatory particles) (28, 29). The 19S proteasome recognizes Ub chains and is required for degradation of ubiquitylated substrates. We transformed p313CUP1ubi3Δub-HA and p313CUP1RPL8Aha plasmids into two 20S proteasome mutants (pre1-1 and pre2-1), two 19S proteasome mutants (cim5-1 and rpn2), and their parental WT strains. pre1-1 and pre2-1 have a mutated β4 and β5 subunit of the 20S proteasome, respectively (30). cim5-1 has a defective Rpt1 ATPase, while rpn2 carries a mutated Rpn2 subunit of the 19S proteasome (31, 32). As expected, CTPD of Rps31ha and Rpl8Aha was inhibited in the pre1-1 and pre2-1 mutants, evidenced by the increase in steady-state levels as compared to the WT counterparts (Fig. 6 A and B, lanes 1 to 3). However, CTPD of Rps31ha and Rpl8Aha was not impaired in the cim5-1 and rpn2 mutants (Fig. 6 A and B, lanes 4 to 6). The defect of 19S proteasome function in cim5-1 and rpn2 was confirmed by inhibition of Ub-dependent degradation of Rpn4_172-229-DHFRha (Fig. 6C). These results suggest that CTPD of Rps31ha and Rpl8Aha may not need the 19S proteasome. This is in line with the observation that CTPD of Rps31ha and Rpl8Aha does not require prior ubiquitylation. We were unable to evaluate CTPD of Rps12ha overexpressed from the GAL1 promoter in the 19S and 20S proteasome mutants because they did not grow or became very sick in the galactose medium.

Fig. 6.

CTPD of Rps31ha and Rpl8Aha is not impaired in 19S proteasome mutants. (A and B) Different requirements of the 20S and 19S proteasomes for CTPD of Rps31ha and Rpl8Aha. Western blotting analysis was performed to compare the steady-state levels of Rps31ha (A) and Rpl8Aha (B) in 20S mutants pre1-1, pre2-1, 19S mutants cim5-1, rpn2, and their parental WT strains. Plasmids p313CUP1ubi3Δub-HA and p313CUP1RPL8Aha were transformed into the yeast strains as indicated. (C) Degradation of Rpn4_172-229-DHFRha is inhibited in the 19S proteasome mutants. Plasmid p313CUP1RPN4(172-229)-DHFRha was transformed into cim5-1, rpn2, and the parental WT strain. Rpn4_172-229-DHFRha protein level was measured by western blotting.

Discussion

In this study, we report several findings: 1) Ribosomal proteins, while posttranslationally stable, are subject to high-level CTPD except for L40 and Rps31, which are synthesized as Ub fusion precursors. 2) Cotranslational folding potential and protein expression level are two critical determinants for CTPD of ribosomal proteins. 3) CTPD of all ribosomal proteins tested in this study is Ub independent.

The role of proteolysis in regulating ribosomal protein homeostasis is understudied. Several studies showed that ribosomal proteins are degraded by the Ub-proteasome system under certain circumstances. Mammalian ribosomal proteins are rapidly degraded during terminal erythroid differentiation and in response to nutrient stress (33, 34). Overexpressed Rpl26 is targeted by the E3 ligase Tom1 in yeast and the homolog Huwe1 in human cells for Ub-dependent degradation (12, 13). It was observed that rapidly accumulated and unassembled ribosomal proteins are degraded by the proteasome in the nucleoplasm (9). Here, we report that ribosomal proteins are subject to substantial CTPD, save for those synthesized as Ub-fusion precursors. This finding opens an avenue to understand the regulation of ribosomal protein homeostasis. We speculate that CTPD may play an important role in regulating the homeostasis of ribosomal proteins from at least two fronts. First, it may gauge the concentrations of ribosomal proteins to prevent accumulation and aggregation of unassembled ribosomal proteins. It is well known that cells often produce ribosomal proteins at high levels beyond the rate for ribosome assembly (4–9). This “wasteful” overproduction likely prepares cells to meet unexpected demand for protein synthesis. Yet, like other RNA-binding proteins, excess free ribosomal proteins tend to form toxic aggregates in the cell (35, 36). This is especially true for ribosomal proteins because they have long half-lives. CTPD can be considered the first-line defense to keep ribosomal proteins in proper concentration. Without CTPD, ribosomal proteins would be accumulated to an extent to which the cell can no longer tolerate. Interestingly, we found that in addition to ribosomal proteins, many other CTPD substrates are relatively abundant in the cell (18). Second, CTPD may help eliminate faulty ribosomal proteins. Errors can arise in ribosomal proteins at both transcription and translation levels. One of the most common mechanisms to generate premature termination codons (PTCs) is cryptic or alternative splicing of introns (37). Introns are rare in Saccharomyces cerevisiae, yet many ribosome genes contain introns. This suggests that ribosome genes have a higher probability to produce PTCs than other genes. Rapid translation of ribosomal mRNAs also increases the chance of producing faulty proteins due to the trade-off between translation speed and accuracy (10, 11). Indeed, our recent work revealed that CTPD preferentially occurs to rapidly translated polypeptides (18). CTPD may serve to remove erroneous nascent chains in a timely manner, preventing the accumulation of misfolded, defective, toxic proteins in the cell. It is worth noting that by far, most studies on CTPD have relied on engineered model substrates produced by mRNAs containing defined lesions. Our study demonstrates that ribosomal proteins are native substrates for CTPD.

CTPD was observed several decades ago, but it remains poorly understood why some nascent proteins are prone to CTPD while others are rather resistant. Our study on ribosomal proteins provides a glimpse of the factors contributing to CTPD. Structural–functional analysis showed that the N-terminal region of Rp8A cannot fold by itself, exposed as a CTPD degron, which is shielded in mature protein likely through interaction with the C-terminal region. Crystal structure analysis of Rpl8A revealed interactions between C-terminal domains and other regions including the N-terminal amino acids 53 to 69 (38). It appears that there is a time window between cotranslational folding and CTPD. If a nascent polypeptide can fold within this time window, it resists CTPD. Otherwise, it is deemed to be attacked by CTPD. For Rpl8A, the N-terminal unstructured segment is relatively far from the C-terminal domain (in primary sequence) and may succumb to CTPD before the C-terminal domain is synthesized. These data suggest that inefficient cotranslational folding contributes to high-level CTPD of Rpl8A. Different from Rpl8A, Rps12 escapes CTPD when expressed at a moderate level. Yet, when expressed at a high level, it is exposed to CTPD. It is likely that nascent Rps12 proteins can fold efficiently at a low expression level. However, overproduced Rps12 proteins may deplete chaperon proteins that are required for their folding. Our data indicate that cotranslational folding efficiency and expression level are two important determinants for CTPD. In support of this argument, we found that addition of an N-terminal Ub reduces CTPD of Rpl8A and overexpressed Rps12. The presence of N-terminal Ub probably allows more time for nascent Rpl8A and Rps12 proteins to fold before being attacked by proteasomes, thereby minimizing CTPD. Early studies also suggested that the N-terminal Ub of Ub-L40 and Ub-Rps31 precursors may act as a chaperone facilitating the production and incorporation of L40 and Rps31 into ribosomes (19, 22–24). Our study defines a more specific role for the N-terminal Ub, i.e., protecting L40 and Rps31 from CTPD. This work also provides an explanation for the observations that expression of recombinant proteins containing an N-terminal Ub often gives rise to a higher yield (39–42).

Another interesting finding of our study is that unlike canonical posttranslational protein degradation, CTPD of all ribosomal proteins tested in this study including Rpl8A, Rps12, and Rps31 is Ub independent. Consistently, CTPD of Rpl8A and Rps31 appears to be independent of the 19S proteasome, which serves as a receptor for Ub chains. (CTPD of Rps12 overexpressed from the GAL1 promoter could not be evaluated in the 19S proteasome mutants because they did not grow in the galactose medium.) In line with our observations, an early report showed that ribosomal proteins were largely absent from the pool of ubiquitylated nascent chains (43). While our results suggest that ubiquitylation does not play a major role in CTPD of ribosomal proteins, it may facilitate CTPD of certain types of substrates such as nascent nonstop proteins produced by translation through the poly(A) tail of mRNAs lacking a stop codon or by stop codon read-through. The stretch of lysine residues translated from the poly(A) binds to the ribosomal exit tunnel and triggers the so-called ribosome-associated protein quality control pathway, leading to dissociation of stalled ribosomes into 40S subunits and 60S ribosome-nascent chain complexes (60S-RNCs) (44–46). The Ltn1 E3 Ub ligase binds to the 60S-RNCs and catalyzes ubiquitylation of the nascent nonstop nascent chains. Ltn1-mediated ubiquitylation may be required to extract the stalling nonstop nascent chains from the 60S subunit for proteasomal degradation. While further investigation is warranted to comprehend the molecular details of CTPD, our study suggests that CTPD may play a pivotal role in regulating the homeostasis of ribosomal proteins.

Materials and Methods

Yeast Strains and Plasmids.

The yeast strains used in this study include SMY113 (MATa UBI1-HA), SMY107 (MATa ubi1Δub-HA), MHY4995 (MATa uba1Δ::KanMX can1-100 leu2-3, 112 his3-11, -15 trp1-1 ura3-1 ade2-1 [pRS313-UBA1-HIS]), MHY4996 (MATa uba1Δ::KanMX can1-100 leu2-3, 112 his3-11, -15 trp1-1 ura3-1 ade2-1 [pRS313-uba1-204-HIS]), WCG4a (MATa ura3 his3-11,15 leu2-3,112), yHI29/1 (MATa pre1-1 ura3 his3-11,15 leu2-3,112), WCG4-21a (MATa pre2-1 ura3 his3-11,15 leu2-3,112), Y791 (MATa cim5-1 ura3-52 his3-Δ20-0 leu2Δ1), JD52 (MATa trp1-Δ63 ura3-52 his3-Δ200 leu2-3,112 lys2-801), JD53 (MATa trp1-Δ63 ura3-52 his3-Δ200 leu2-3,112 lys2-801), YXY270 (a cim5-1 derivative from tetrad dissection of JD53/Y791 diploid), and AVY302 (an rpn2 mutant derived from JD52, see ref. 32). SMY113 and SMY107 were provided by Jesus de la Cruz (see ref. 22). MHY4995 and MHY4996 were obtained from Mark Hochstrasser (see ref. 26). WCG4a, yHI29/1, and WCG4-21a were described previously (30). All plasmids used in this study are listed in SI Appendix, Table S2. All constructs were verified by restriction enzyme analysis and DNA sequencing.

Yeast Growth Conditions and Induction of Gene Expression.

Yeast cells transformed with CUP1 promoter-based expression vectors were grown in synthetic minimal (SD) medium supplemented with essential amino acids. Cells grown to log phase were diluted to OD₆₀₀ of ~0.15 and subjected to induction with CuSO₄ (0.1 mM) for 7 h before being harvested. For GAL1 promoter-driven gene expression, yeast transformants were grown in synthetic medium with raffinose as the carbon source for overnight. The overnight cultures were then diluted to OD₆₀₀ of ~0.15 into synthetic medium with galactose as the carbon source. Galactose induction lasted for 7 h before harvested. For experiments measuring the effect of loss of E1 activity on protein degradation, WT (MHY4995) and uba1-204 (MHY4996) transformants were first induced with CuSO₄ (0.1 mM) or galactose at permissive temperature (28 °C) for 1 h. Half of each culture was then shifted to nonpermissive temperature (37 °C) for additional 2 h induction, whereas the other half was continuously induced at permissive temperature.

Western Blotting.

Cell pellets were lysed in 2× SDS buffer (2% SDS, 30 mΜ dithiothreitol, and 90 mΜ Na-HEPES, pH 7.5) by incubation at 100 °C for 5 min. Supernatants were recovered by centrifugation and applied to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). Separated proteins were transferred to a Polyvinylidene fluoride (PVDF) membrane for western blotting analysis. Unless stated otherwise, the blots were probed simultaneously with anti-ha mouse monoclonal antibody (Covance) for ha-tagged proteins and anti-Pgk1 rabbit polyclonal antibody (Origene) for the house-keeping enzyme phosphoglycerate kinase (Pgk1), which served as an internal control to ensure equal input of cell extracts. After unbound antibodies were removed, the blots were incubated simultaneously with fluorescent dye (Alexa Fluor 680)-conjugated goat anti-mouse and donkey anti-rabbit secondary antibodies, followed by detection with the Odyssey infrared imaging system (Li-Cor Biosciences). To detect ha-tagged proteins that overlap or migrate very close to the Pgk1 control, the blots were first probed with anti-ha antibody and Alexa Fluor 680–conjugated goat-anti-mouse secondary antibody. Subsequently, the blots were probed with anti-Pgk1 antibody and Alexa Fluor 800–conjugated donkey anti-rabbit secondary antibody. The images of ha-tagged proteins and Pgk1 were separated by detection channels with different wavelengths.

CHX-Chase Analysis.

Exponentially growing cells were harvested and resuspended in the same medium supplemented with CHX (0.2 mg/mL) and chased for various time periods. Equal volume of sample was withdrawn at each time point. Cells were pelleted and lysed with 2× SDS buffer. Cell extracts were analyzed by western blotting.

Pulse Labeling Assay.

Exponentially growing cells in SD medium containing essential amino acids were harvested and resuspended in the same medium supplemented with 0.15 mCi of ³⁵S-Met/Cys (PerkinElmer) for pulse labeling. Cells were then pelleted and washed with cold H₂O to remove residual medium containing ³⁵S-Met/Cys. Labeled cells were lysed in 2× SDS buffer. Supernatants were recovered by centrifugation and diluted 20-fold with buffer A (150 mM NaCl, 1 mM ethylenediaminetetraacetic acid (EDTA), 50 mM Na–HEPES, pH 7.5, and 1% Triton X-100) before being subjected to immunoprecipitation (IP) with anti-ha antibody. The precipitated proteins were separated by SDS-PAGE, followed by autoradiography. The volumes of supernatants applied to IP were adjusted to equalize the amounts of ³⁵S incorporated into total proteins using the trichloroacetic acid (TCA) precipitation assay as previously described (47).

qRT-PCR Analysis.

The procedure of RNA isolation from yeast cells was described previously (48). Reverse transcription was conducted with SuperScript III reverse transcriptase (Invitrogen), and qRT-PCR was performed using Fast SYBR Green Master Mix in the StepOne™ system following the manufacturer’s instruction (Applied Biosystems). Experiments were performed in triplicate and repeated three times. Data were analyzed by Student’s t test. The ACT1 gene served as an internal control for normalization of gene expression levels. The oligo primers for qRT-PCR are listed in SI Appendix, Table S3.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

The data supporting the findings of this study are in the article and/or SI Appendix.